Several important applications, including spin-valve giant magneto-resistive (GMR) thin-film heads and semiconductor integrated circuits, use multi-layer material stacks to perform various electronic signal processing and data storage functions. For instance, semiconductor integrated circuit (IC) applications often include multi-layer interconnect structures comprised of multiple layers of glue/diffusion barrier, interconnect metal, and anti-reflection coating (ARC) films. For instance, some multi-level interconnect structures in semiconductor ICs employ a multi-layer conductive material stack comprising titanium, titanium nitride, aluminum (doped with copper), and a top titanium nitride ARC layer in each interconnect level. Another application that uses multi-layer material structures is magnetic data storage thin-film head devices. For instance, giant magneto-resistive (GMR) thin-film head and magnetic random access memory (MRAM) spin-valve tunnel junction devices use multi-layer material structures comprising stacks of conductive, magnetic, and/or insulating material layers as thin as 10 to 30 .ANG..
Conventional magnetic data storage devices use thin film heads comprised of inductive and/or magneto-resistive (MR) materials. MR heads enable higher magnetic storage densities compared to the storage densities of devices having inductive heads due to the higher read sensitivity and signal-to-noise ratio of MR senors. The MR heads read the stored information with direct magnetic flux sensing and are, thus, capable of static read-back without dependency on the relative motion (e.g., disk rotation speed) of the magnetic media compared to the head. The MR heads operate based on a resistance change of an MR element (permalloy) in response to the magnetic flux on the media. Both the inductive and MR thin-film heads employ inductive writer elements.
Industry transition from inductive heads to MR heads for magnetic data storage systems has allowed rapid technology evolution in terms of maximum storage density (described in Gbits/in.sup.2) and system storage capacities. Industry has increased storage density of magnetic storage systems at a historical rate of 30% per year and a current annual rate of 60%. Leading edge state-of-the-art rigid disk storage media now have storage densities on the order of 2 to 5 Gbits/in.sup.2 (gigabits per square inch), with industry projecting storage densities approaching 10 Gbits/in.sup.2 by the turn of the century. As the recording densities transition from 2 Gbits/in.sup.2 towards 5 Gbits/in.sup.2, industry will have to replace the MR head technology with more sensitive devices, such as spin-valve GMR heads. Eventually, to maintain present trends toward improved storage capacities, industry may transition from GMR materials to colossal magneto-resistive (CMR) materials, which could support storage densities approaching 100 Gbits/in.sup.2.
In 1987, the giant magneto-resistive or GMR effect was discovered. GMR materials, usually consisting of at least two ferromagnetic nanostructure entities separated by a nonmagnetic spacer, display a change of resistance upon the application of a magnetic field. GMR materials have a larger relative resistance change and have increased field sensitivity as compared against traditional anisotropic magneto-resistive or MR materials, such as Ni.sub.80 Fe.sub.20 films. The improved relative resistance change and field sensitivity of GMR materials and related magnetic sensing elements allow the production of sensors having greater sensitivity and signal-to-noise ratio than conventional sensors. Thus, for instance, data storage systems using GMR read sensors can store greater amounts of data in smaller disk areas as compared to conventional data storage devices. However, material stacks for fabricating GMR sensors generally use 6 to 8 layers of 4 to 6 different materials, as compared to the MR material stacks, which usually have only 3 layers of materials such as permalloy layers with Soft Adjacent Layers (SAL). Thus, creating material stacks for GMR read sensors generally requires more processing steps, including more complicated equipment and fabrication techniques for high-yield manufacturing of high-performance GMR thin-film heads.
In order to meet its goals for improved storage density, industry will likely turn to spin-valve GMR thin-film heads. Spin-valve GMR heads are comprised of multi-layer depositions of 10 to 100 angstrom thick material films having precise thickness and microstructure control as well as extremely cohesive interface control at each interface of a multi-layer spin-valve GMR stack. Each spin-valve GMR stack must have good crystalinity in conjunction with abrupt and smooth material interfaces with minimal interface mixing to ensure proper GMR response and to establish excellent thermal stability. For instance, FIG. 1 depicts one possible spin-valve GMR configuration. The precision required for spin-valve stack deposition can be understood by comparing the 1.5 nanometer thick layer of cobalt in FIG. 1 against a typical atomic radius of 0.2 nanometers (corresponding to approximately 7 atomic layers). Essentially, GMR stacks may require controlled deposition of metallic multilayers which comprise ultrathin films as thin as 5 to 10 atomic nanolayers.
Another application for GMR materials is magnetic random access memories ("MRAM"), which are monolithic silicon-based nonvolatile memory devices presently based on a hysteretic effect in magneto-resistive or MR materials. MRAM devices are typically used in aerospace and military applications due to their excellent nonvolatile memory bit retention and radiation hardness behavior. Moreover, the MRAM devices can be easily integrated with silicon integrated circuits for embedded memory applications. The implementation of GMR materials, such as spin-dependent tunnel junctions, could improve the electrical performance of MRAM devices to make MRAM devices competitive with semiconductor DRAM and flash EPROM memory devices. However, the performance of MRAM memory depends on precise control of layer thickness values and the microstructures of various thin films in a GMR stack of thin metallic films. Thickness fluctuations and other interface or microstructural variations in thin metallic layers can cause variation in MRAM device performance. Similar difficulties can occur with periodic laminated multi-layer structures, such as laminated flux guide structures of iron, tantalum and silicon di-oxide.
The precision-controlled deposition of materials onto a substrate to create the multi-layer structures that can use the GMR effect is a difficult and time consuming process which requires high-performance vacuum deposition equipment, including plasma sputtering, ion-beam and evaporation processes. Although conventional physical-vapor deposition (PVD) technology can create GMR-capable structures, each layer of a structure must be carefully deposited in sequence in a time-consuming sequential series of depositions, a complicated process having a relatively slow throughput. Typically, such conventional PVD technology dynamically rotates a substrate at rapid speeds relative to a target in an attempt to evenly distribute the material being deposited onto the substrate. However, dynamic deposition requires a relatively large process chamber relative to the size of the target and the size of the substrate in order to allow rotation of the substrate. The PVD systems with dynamic rotation also complicate integration of advanced chucks and/or magnetic orientation devices for substrate processing applications. Further, dynamic deposition is inefficient because the target deposits material onto the substrate only when the rotation of the substrate aligns it partially or fully with the target. Material deposited from the target during non-alignment periods is wasted. Also, precise control of layer thickness and interface characteristics cannot be ensured with dynamic deposition, particularly when targets are changed after each dynamic deposition process or substrates are moved to modules with new targets, thus, allowing impurities to be introduced between deposition layers. Such impurities frequently cause material structures to fail.